| Pullulan is a simple and widely used linear polysaccharide that has been used as a food additive and medicinal excipient.The weight average molecular weight of pullulan has an impact on fermentation yield,physicochemical properties,and application range.Therefore,understanding the molecular weight and content changes of pullulan during fermentation and monitoring them in real-time is of great significance for the customized production of specific molecular weights.This research uses near-infrared spectroscopy(NIRS)to carry out in-line analysis of molecular weight and content changes of pullulan,in order to reveal the molecular weight changes during pullulan fermentation and to achieve in-line in situ real-time monitoring of its content and molecular weight,so as to solve the "black box" phenomenon commonly existing in the fermentation process and to achieve the "transparency" of the quality of the fermentation process.The main content of the research in this project will be as follows:(1)Research on small-scale fermentation process optimization of pullulan based on NIRS.In this study,pullulan fermentation experiments were conducted under different fermentation conditions,and the near-infrared spectra of the fermented products under different process conditions were collected using a near-infrared spectrometer,and their content,molecular weight,pH,and other related chemical and physical indexes were measured.A rapid quantitative model for the key quality parameters of pullulan content and molecular weight was established using the partial least squares regression(PLSR)method,with RMSEP values of 0.6657 g/L and 3244462.9347 Da,and RPD values of 3.2910 and 2.5141,respectively.The results showed that NIRS provides a fast detection method for the determination of the results of small-scale fermentation process optimization of pullulan,effectively solving the problems of time-consuming,labor-intensive,and untimely feedback of key quality parameters in the optimization process of pullulan fermentation.(2)Research on the establishment of off-line model for key quality attributes of pullulan fermentation process based on NIRS.In this study,the optimized culture medium obtained in the first part was used to conduct pullulan fermentation experiments in a fermenter to establish off-line models for the content and molecular weight of pullulan during fermentation.The study conducted 4 fermentation batches,and a total of 252 fermented liquid samples were obtained during the fermentation process,and the off-line near-infrared spectra of the fermented liquid samples were collected.Three modeling methods,PLSR,random forest(RF)regression,and back propagation-artificial neural network(BP-ANN),were used to establish NIRS quantitative analysis models,and the results of the three models were compared.The PLSR method was selected as the most appropriate modeling method,with RPD values of 7.1433 and 4.2813 for the pullulan content and molecular weight PLSR models,respectively.The Pearson correlation coefficient between each fermentation parameter and modeling index was calculated to explain the relationship between the contents of each component in the fermented liquid,and the multi-source data fusion models were established,with an RPD value of 7.4513 and 4.2821.The results showed that the performance of the fusion models was better,proving that the multi-source data fusion strategy can be used to optimize the quantitative models in the fermentation process.At the same time,this study plays an important role in understanding the fermentation process and accurately and quickly determining the content and molecular weight of pullulan.(3)Research on the establishment of in-line model for key quality attributes of pullulan fermentation process.In this study,a custom-made submersible diffuse reflectance probe and an in-line nearinfrared spectrometer were used to set up an in-line NIRS monitoring system for the pullulan fermentation process,and in-line NIRS quantitative analysis models were established for the pullulan content and molecular weight,with RPD values of 6.7850 and 4.2374,respectively.At the same time,the spectra data were integrated with fermentation parameters such as temperature,pH value,and stirring speed,and the multi-source data fusion models were established,with RPD values of 7.1928 and 4.4283 for the fusion models of pullulan content and molecular weight,respectively.In addition,an in-line monitoring model for the content of sucrose was established to monitor the fermentation substrate,with RMSEP and RPD values of 3.2249 g/L and 3.6041,respectively.The results show that NIRS can be applied to in-line monitoring of the fermentation process,and the prediction results of various models have high accuracy and reliability,which is of great importance for improving the production efficiency and product quality of the fermentation process. |
PDF Full Text Request